[visad] Missing Data and Memory Usage

[Don Murray]

Hi Kevin-

Kevin Manross wrote:
> Happy New Year, Y'all!

Same to you.

> I'm working with trying to view some radar data in the IDV.  Without 
> getting into a great amount of detail, I seem to be taking a 
> considerable memory hit with a data set that is about 75% "missing data" 
> (i.e., a 90 degree sector scan, where the other 270 degrees are produced 
> but set to "missing data").  We have the missing data flag set in our 
> netCDF files and indeed those points are not being displayed. 
> 
> What I would like to know is whether those points marked as "missing 
> data" are still held in memory.  If so, we'll examine whether we can 
> create a "uniform" sector scan for our data.
> 
> I'm guessing this is at the VisAD level, hence the post to this group. 
> 
> Thanks a bunch!!

The missing values will be set to NaN, and if your dimensions
(e.g. range, azimuth) include the region where there is missing
data, then the domain created will also include them.  So,
yes, they will be held in memory as NaN's.

Don
*************************************************************
Don Murray                               UCAR Unidata Program
dmurray at unidata.ucar.edu                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307
http://www.unidata.ucar.edu/staff/donm
*************************************************************